Sender rotation plans

    Pick which senders stay in rotation without a spreadsheet. Your agent reads sender health from your local database and returns a tiered plan with reasons.

    Legacy note: Sections below that mention MCP tokens and mcp.outreachmagic.io describe the previous hosted product. Today's Outreach Magic works the same way in practice: install the skill, sync your sequencers, and ask your agent for a sender rotation plan. See Get started and agent prompts.

    Current product (recommended)

    1. 1Install the skill and connect at least one sequencer webhook.
    2. 2Configure sender accounts so health metrics sync to your local SQLite file. See Configure senders.
    3. 3Paste this prompt in Claude Code, Cursor, or Hermes. Replace the workspace name.
    Sender rotation prompt
    Use the Outreach Magic skill. Build a sender rotation plan for workspace [workspace-name].

Read sender health from my local SQLite database. Sync first if data looks stale. Do not invent metrics.

Return:
- Tier A, B, backup, and excluded senders with scores and reply/bounce rates
- A short reason for each exclusion
- Export the plan as markdown I can share with my team

    Your agent syncs the database if needed, scores each sender, and returns Tier A, B, backup, and excluded accounts with reasons. Same output shape as the legacy MCP flow.

    What you get

    • • Tier A: top performers by reply rate, bounce rate, warmup health, and domain capacity
    • • Tier B: solid senders with smaller sample sizes or slightly lower scores
    • • Backup: available but deprioritized
    • • Excluded: with a specific reason (low warmup, bounce spike, stale sync, domain cap, and so on)

    The plan uses:

    • • Sends, replies, and bounces over your lookback window
    • • Warmup score and daily send limits
    • • Domain capacity (how many senders per domain can stay in Tier A)
    • • Workspace-level baselines so senders compare to relevant peers

    How scoring works

    • • Reply weight: weighted heavily vs other signals
    • • Bounce penalty: spikes flag immediate exclusion
    • • Warmup health: low scores get deprioritized
    • • Sample size: low-volume senders get a confidence discount
    • • Domain capacity: caps accounts per domain in Tier A

    Ask your agent to adjust weights or the lookback window in the same conversation.

    Tips

    Run weekly. Sender health changes; last week's Tier A may not stay Tier A.

    Ask for audit detail. Add "show scoring breakdown per sender" when you need to explain rotation decisions to a client.

    Sync first. If numbers look stale, tell your agent to sync the database before building the plan.

    Legacy: MCP setup (previous hosted product)

    If you still use the hosted MCP connector, these steps apply. New installs should use the skill prompt above instead.

    Prerequisites

    • • Pro workspace with sender health sync enabled
    • • MCP enabled for the workspace
    • • Claude Desktop, Cursor, or Windsurf connected to Outreach Magic MCP

    Step 1: Copy your MCP token

    In the legacy portal: Settings → MCP. Copy your workspace-scoped token.

    Step 2: Connect Claude or Cursor

    Claude: Settings → Connectors → Add custom connector → paste:

    https://mcp.outreachmagic.io?token=YOUR_MCP_TOKEN

    Cursor: Add under mcpServers (replace token and workspace label):

    {
  "mcpServers": {
    "om-your-workspace": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://mcp.outreachmagic.io?token=YOUR_MCP_TOKEN"
      ]
    }
  }
}

    Related